The Enterprise Security Blind Spot in Agentic AI | Token Security’s Itamar Apelblat

Why It Matters

Without securing AI agents as distinct identities, organizations risk uncontrolled access and data breaches, making intent‑based identity management essential for safe AI adoption.

Key Takeaways

• AI agents create a new class of non‑human identities.
• Token Security maps agent access to business applications for policy enforcement.
• Lack of visibility into service accounts exposes critical infrastructure risks.
• RSA Innovation Sandbox finalists receive $5 M venture funding each.
• Secure AI adoption requires intent‑based controls and lifecycle management.

Summary

The video features Itamar Apelblat, CEO and co‑founder of Token Security, discussing the emerging security blind spot created by agentic AI at RSA’s Innovation Sandbox. He frames AI agents as a new, non‑human identity class that traditional IAM solutions don’t cover, highlighting the rapid shift from human‑centric identity to machine‑centric and agent‑centric models.

Apelblat explains that Token Security tackles this gap by integrating with both enterprise applications and AI platforms, mapping every agent’s permissions, actions, owners, and intent. By correlating agent activity with business‑critical resources, the platform can enforce granular policies—e.g., restricting a specific AI model to a single department or sandbox environment—without adding latency to AI adoption.

A memorable anecdote illustrates the problem: a six‑year‑old service‑account password gave Apelblat admin access to a highly secure network, exposing how forgotten non‑human credentials can become catastrophic. He emphasizes that “agent risk is based on the level of access they have and the autonomy they possess,” and that understanding an agent’s purpose is essential for effective control.

The discussion underscores that enterprises must adopt intent‑based identity controls and lifecycle management to safely harness AI’s productivity gains. Token’s finalist status in RSA’s Innovation Sandbox, backed by a $5 M venture pool, signals strong market validation for solutions that secure the next generation of AI‑driven operations.

Original Description

Techstrong CEO Alan Shimel speaks with Token Security CEO and co-founder Itamar Apelblat ahead of the RSA Innovation Sandbox about one of the biggest emerging security challenges in enterprise AI: managing the identities and privileges of autonomous agents.

In this interview, Apelblat explains why the rise of agentic AI is creating a new blind spot for security teams. While enterprises have spent years refining identity and access management for human users, non-human identities tied to AI agents, services and applications are introducing a new layer of complexity and risk.

Shimel and Apelblat explore how Token Security is helping organizations map non-human identities directly to business applications, understand how those identities behave and build the guardrails needed to reduce risk without slowing AI adoption. As more enterprises deploy autonomous systems into critical workflows, visibility and control over agent privileges are becoming essential.

Watch this interview for a closer look at one of the most important security issues emerging alongside the growth of agentic AI.

#AI #AgenticAI #Cybersecurity #IdentitySecurity #RSAC #InnovationSandbox #TechstrongTV